Autophagy-mediated Regulation of BACE1 Protein Trafficking and Degradation*

β-Site amyloid precursor protein (APP) cleaving enzyme 1 (BACE1) is the major neuronal β-secretase for amyloid-β generation and is degraded in lysosomes. The autophagy-lysosomal system plays a key role in the maintenance of cellular homeostasis in neurons. Recent studies established that nascent autophagosomes in distal axons move predominantly in the retrograde direction toward the soma, where mature lysosomes are mainly located. However, it remains unknown whether autophagy plays a critical role in regulation of BACE1 trafficking and degradation. Here, we report that induction of neuronal autophagy enhances BACE1 turnover, which is suppressed by lysosomal inhibition. A significant portion of BACE1 is recruited to the autophagy pathway and co-migrates robustly with autophagic vacuoles along axons. Moreover, we reveal that autophagic vacuole-associated BACE1 is accumulated in the distal axon of Alzheimer's disease-related mutant human APP transgenic neurons and mouse brains. Inducing autophagy in mutant human APP neurons augments autophagic retention of BACE1 in distal axons, leading to enhanced β-cleavage of APP. This phenotype can be reversed by Snapin-enhanced retrograde transport, which facilitates BACE1 trafficking to lysosomes for degradation. Therefore, our study provides new insights into autophagy-mediated regulation of BACE1 turnover and APP processing, thus building a foundation for future development of potential Alzheimer's disease therapeutic strategies.

␤-Site amyloid precursor protein (APP) cleaving enzyme 1 (BACE1) is the major neuronal ␤-secretase for amyloid-␤ generation and is degraded in lysosomes. The autophagy-lysosomal system plays a key role in the maintenance of cellular homeostasis in neurons. Recent studies established that nascent autophagosomes in distal axons move predominantly in the retrograde direction toward the soma, where mature lysosomes are mainly located. However, it remains unknown whether autophagy plays a critical role in regulation of BACE1 trafficking and degradation. Here, we report that induction of neuronal autophagy enhances BACE1 turnover, which is suppressed by lysosomal inhibition. A significant portion of BACE1 is recruited to the autophagy pathway and co-migrates robustly with autophagic vacuoles along axons. Moreover, we reveal that autophagic vacuole-associated BACE1 is accumulated in the distal axon of Alzheimer's disease-related mutant human APP transgenic neurons and mouse brains. Inducing autophagy in mutant human APP neurons augments autophagic retention of BACE1 in distal axons, leading to enhanced ␤-cleavage of APP. This phenotype can be reversed by Snapin-enhanced retrograde transport, which facilitates BACE1 trafficking to lysosomes for degradation. Therefore, our study provides new insights into autophagy-mediated regulation of BACE1 turnover and APP processing, thus building a foundation for future development of potential Alzheimer's disease therapeutic strategies.
Autophagy is the major cellular degradation pathway for long-lived proteins and damaged organelles (25)(26)(27)(28). Altered autophagy has been linked to several major age-related neurodegenerative diseases, including AD (27). Recent studies established that autophagosomes are continuously generated in distal axons and move predominantly in the retrograde direction toward the soma for lysosomal proteolysis (23, 29 -33). Such retrograde transport is initiated by fusion of nascent autophagosomes with LEs, forming amphisomes, and is driven by LEloaded dynein-Snapin (motor adaptor) complexes (32). This raises a fundamental question as to whether autophagosomes also serve as cargoes carrying BACE1 trafficking and thus regulate its turnover and, if so, whether AD-associated autophagic stress and altered trafficking of autophagosomes augment BACE1 accumulation in distal axons.
In this study, we provide evidence that autophagy induction enhances BACE1 turnover in primary cortical neurons. A significant portion of BACE1 is recruited to and co-migrates with autophagic vacuoles (AVs) toward the soma, which facilitates BACE1 trafficking to lysosomes for degradation. Moreover, we provide in vitro and in vivo evidence that AV-associated BACE1 is aberrantly accumulated in distal axons along with impaired retrograde transport in neurons of mutant hAPP J20 transgenic (Tg) mice. Autophagy activation augments BACE1 accumulation in the axon of mutant hAPP Tg neurons, leading to enhanced ␤-cleavage of APP. Conversely, enhancing retrograde transport by overexpressing the dynein adaptor Snapin reduces accumulation of AV-associated BACE1 and facilitates BACE1 delivery to lysosomes. Therefore, our study provides new insights into how neuronal autophagy regulates BACE1 trafficking and turnover through AV retrograde transport and how impaired AV transport augments BACE1 retention in AD axons, thereby enhancing APP amyloidogenic processing.

Results
Autophagy Activation Enhances BACE1 Turnover-To determine whether BACE1 turnover is regulated by autophagy, cultured primary cortical neurons were treated with trehalose, which induces neuronal autophagy in an mTOR-independent fashion (34). Trehalose treatment led to a significant reduction of BACE1 levels (0.528 Ϯ 0.01; p Ͻ 0.001), but a substantial increase of BACE1 was detected in the presence of lysosomal inhibitors (2.12 Ϯ 0.66; p ϭ 0.0365) (Fig. 1, A and B), suggesting that lysosomal inhibition suppressed autophagy-mediated BACE1 turnover. To further evaluate the effect of autophagy induction on BACE1 turnover, we utilized alterative drug 10-NCP, another mTOR-independent autophagy inducer (36). Consistent with trehalose treatment, BACE1 levels were reduced to ϳ64% following incubation with 10-NCP for 24 h. Lysosomal inhibition abolished 10-NCP-induced effects on BACE1 turnover, leading to its accumulation (1.94 Ϯ 0.31; p ϭ 0.004) (Fig. 1, D and E). Following a 24-h treatment with trehalose or 10-NCP, we found that LC3-II levels were consistently elevated (trehalose, 2.06 Ϯ 0.195; p Ͻ 0.01; 10-NCP, 1.96 Ϯ 0.21; p Ͻ 0.01), suggesting an increased autophagic flux. Lysosomal inhibition resulted in an additional increase of LC3-II (trehalose, 3.27 Ϯ 0.33; p Ͻ 0.001; 10-NCP, 2.83 Ϯ 0.27; p Ͻ 0.001), which is attributed to the blockage of lysosomal proteolysis of autophagic cargoes (Fig. 1, A, C, D, and F). As an internal control, we did not detect any change of syntaxin 1, a neu-ronal marker (35). Moreover, we tested the effect of activating mTOR-dependent autophagy by treating primary cortical neurons with the mTOR inhibitor rapamycin. We detected similar effects; the levels of BACE1 were decreased in neurons incubated with rapamycin (data not shown). Quantitative analysis indicated ϳ25% reduction of BACE1 levels, suggesting that mTOR-dependent autophagy facilitates BACE1 turnover. These observations indicate that the autophagy-lysosomal pathway can effectively regulate BACE1 turnover.
Association of BACE1 with Autophagosomes and Robust Movement toward the Soma upon Autophagy Induction-Recent studies demonstrated that autophagosomes are predominantly generated in distal axons and undergo retrograde transport for lysosomal proteolysis in the soma (32,37). We next asked whether autophagy induction facilitates BACE1 turnover by enhancing their retrograde transport. We examined the distribution and transport of axonal BACE1 and autophagosomes in live neurons transfected with BACE1-GFP and the autophagy marker mRFP-LC3 in the presence and absence of 100 mM trehalose. At the basal condition, a majority of mRFP-LC3 was diffused as the form of cytosolic LC3-I. We detected a few lipidated LC3-II-labeled AVs, which predominantly moved toward the soma ( Fig. 2A). Trehalose incubation markedly increased the density of AVs along axons (number of AVs per 10 m length: basal condition, 0.46 Ϯ 0.04; trehalose, 1.40 Ϯ 0.10; p Ͻ 0.001) (Fig. 2, A and B). This result is consistent with our biochemical analysis showing increased LC3-II levels ( Fig. 1, A and C), suggesting an increased autophagic flux. Surprisingly, a significant portion of BACE1 was recruited to the autophagy pathway after the treatment with trehalose. The density of AV-associated BACE1 was robustly elevated relative to that in basal conditions (per 10 m length: basal condition, 0.32 Ϯ 0.04; trehalose, 1.06 Ϯ 0.07; p Ͻ 0.001) (Fig. 2, A and C). Under basal conditions, ϳ15% of BACE1 was co-localized and co-migrated with AVs along the same axon. However, it was significantly increased following autophagy activation (41.02 Ϯ 3.95%; p Ͻ 0.001) (Fig. 2, A and D). These newly formed AVs displayed predominant retrograde transport along axons (57.57 Ϯ 3.73%) (Fig. 2, A and E). More importantly, retrograde transport of AV-associated BACE1 along the same axons was markedly increased (basal condition, 37.02 Ϯ 5.02%; trehalose, 60.72 Ϯ 4.41%; p Ͻ 0.001) (Fig. 2, A and F). Therefore, our results suggest that autophagy induction promotes the recruitment of BACE1 to the autophagy pathway and enhances BACE1 retrograde transport from distal axons.
Accumulating evidence demonstrates that mature lysosomes are mainly located in the soma of neurons (21)(22)(23)(24). We and others reported that BACE1 relies on lysosomal degradation (15, 18 -20). Given that autophagy activation enhanced retrograde transport of AV-associated BACE1 (Fig. 2, A and F), we sought to determine whether BACE1 trafficking to lysosomes could also be enhanced under the same conditions. After a 24-h incubation with trehalose, we found that co-localized intensity of BACE1 with lysosomes, labeled by LAMP-1, was significantly increased in the soma compared with that of control condition (1.62 Ϯ 0.08; p Ͻ 0.001) (Fig. 3, C and D). Thus, by enhancing retrograde transport, autophagy activation facilitates the trafficking of BACE1 to somatic lysosomes for degradation, as reflected by reduced BACE1 levels (Fig. 1). Consistently, the density of autolysosomes in the soma was markedly increased upon autophagy induction (basal, 1.55 Ϯ 0.22; trehalose, 6.56 Ϯ 1.27; p Ͻ 0.001) (Fig. 3, A and B). Our data are consistent with a recent study that axon-generated autophagosomes enter the soma and then remain in the somatodendritic domain, where mature lysosomes are mainly located (33).
Impaired Retrograde Transport Accumulates AV-associated BACE1 in Axons of Mutant hAPP Neurons-Altered autophagy has been implicated in AD. AD neurons exhibit massive accumulation of AVs within dystrophic neurites (38), highlighting impaired autophagic transport and clearance. We next determined whether such AD-associated autophagic stress augments BACE1 retention in distal axons. Cortical neurons derived from mutant hAPP Tg (J20) mice and their WT littermates were transfected with BACE1-GFP and mRFP-LC3, followed by time-lapse live cell imaging at DIV 17-19. Surprisingly, compared with WT neurons, the density of AVs was increased in mutant hAPP neurons without any treatment (per 10-m length: WT, 0.47 Ϯ 0.046; hAPP, 0.85 Ϯ 0.056; p Ͻ 0.001) (Fig. 4, A and B). A large portion of BACE1 was retained within AVs in the axon of hAPP neurons (per 10-m length: WT, 0.32 Ϯ 0.036; hAPP, 0.60 Ϯ 0.045; p Ͻ 0.001) (Fig. 4, A and C). In WT neurons, ϳ14.49% of BACE1 was co-localized with axonal AVs; however, the percentage of BACE1 associated with axonal AVs was significantly increased (39.19 Ϯ 3.16%; p Ͻ 0.001) (Fig. 4, A and D). These data indicate that BACE1 was aberrantly accumulated within the autophagy pathway in the axon of mutant hAPP Tg neurons.
We then assessed the retrograde motility of AVs and AVassociated BACE1 in live mutant hAPP Tg axons. Both AVs and BACE1 associated with AVs displayed similar reduction in ret- Cultured cortical neurons were transfected with BACE1-GFP and mRFP-LC3, followed by a 24-h incubation with 100 mM trehalose (Tre) prior to time-lapse imaging at DIV 10 -12. The axon image was taken ϳ100 m away from the cell body. Whereas BACE1-GFP displayed bi-directional movement, mRFP-LC3 predominantly moved toward the soma of neurons. Note that a significant portion of BACE1 co-localized and co-migrated with LC3-labeled AVs along the same axons toward the soma in the presence of trehalose. B-D, inducing autophagy facilitates BACE1 trafficking along the autophagy pathway. Note that the densities of axonal AVs and BACE1 associated with AVs and the percentage of autophagic BACE1 relative to total BACE1 were increased in response to autophagy activation. The average numbers of AVs and AVs co-labeled with BACE1 per 10-m axon were quantified in neurons with and without trehalose treatment, respectively (B and C). E and F, autophagy induction enhanced retrograde transport of AV-associated BACE1 in axons. Relative motility of axonal AVs and AV-associated BACE1 in the presence and absence of trehalose was measured. Vertical lines represent stationary organelles; oblique lines or curves to the right (negative slope) represent anterograde movements; and lines to the left (positive slope) indicate retrograde transport. These data were quantified from at least four independent repeats and from a total number of AVs or AV-associated BACE1 (v) from a total number of neurons (n), as indicated in parentheses (B-F). rograde motility in the same distal axon of mutant hAPP neurons (AVs, 20.11 Ϯ 0.02%; p ϭ 0.00012; autophagic BACE1, 19.56 Ϯ 2.75%; p ϭ 0.0076) relative to their retrograde motility in WT neurons (AVs, 36.89 Ϯ 3.56%; autophagic BACE1, 35.48 Ϯ 5.32%) (Fig. 4, A, E, and F). However, such reduction was not found in anterograde transport (AVs, p ϭ 0.2789; AV-associated BACE1, p ϭ 0.1861), suggesting that BACE1 retention in mutant hAPP axons is attributed to impaired retrograde transport of AVs.
Autophagic Retention of BACE1 in the Distal Axon of Mutant hAPP Tg Mouse Brains-To provide in vivo evidence, we performed co-immunostaining of mutant hAPP Tg (J20) mouse  . Impaired retrograde transport accumulates AV-associated BACE1 in axons of mutant hAPP neurons. A, representative dual channel kymographs showing impaired retrograde transport of AVs and AV-associated BACE1 in the axon of mutant hAPP Tg neurons. Note that LC3-marked AVs and BACE1 co-localized with AVs mainly moved toward the soma of WT neurons, whereas a majority of AVs and BACE1 remained stationary in the axon of mutant hAPP neurons. B-D, BACE1 was retained within AVs in the axon of hAPP neurons. Note that the densities of both axonal AVs and AV-associated BACE1, and BACE1 retention within AVs were increased in mutant hAPP axons. The average numbers of AVs and BACE1 co-labeled with LC3 per 10 m axon were quantified in the same axon of WT and mutant hAPP neurons, respectively (B and C). The percentage of autophagic BACE1 relative to total BACE1 was examined in the axon of WT and hAPP neurons (D). E and F, impaired retrograde transport of AVs and AV-associated BACE1 in the axon of mutant hAPP neurons. Note that AVs and BACE1 share similarly reduced retrograde motility in the same mutant hAPP axons. Relative motility of axonal AVs and BACE1 co-labeled with LC3 in WT and mutant hAPP neurons was measured. Data were quantified from four independent repeats of four pairs of mice. Data were quantified from the total number of axonal AVs or autophagic BACE1 (v) and from the total number of neurons (n), as indicated in parentheses (B-F). Scale bars, 10 m. Error bars, S.E. Student's t test. (***, p Ͻ 0.001; **, p Ͻ 0.01; *, p Ͻ 0.05.) brains with antibodies against BACE1 and autophagy marker p62 or ubiquitin. BACE1 appeared as relatively diffused in the hippocampal mossy fiber processes of WT mouse brains, whereas BACE1 was clustered and accumulated in mutant hAPP Tg (J20) mouse brains (Fig. 5, A and B). The average number of BACE1 clusters per slice section was substantially increased relative to that of WT mouse brains (WT, 26.06 Ϯ 0.91; mutant hAPP Tg, 78.57 Ϯ 1.61; p Ͻ 0.001) (Fig. 5C). Moreover, the majority of BACE1 co-localized with p62 or ubiquitinlabeled AVs in the hippocampal regions of mutant hAPP mice (p62, 89.52 Ϯ 0.81%; ubiquitin, 84.85 Ϯ 0.55%) (Fig. 5, A, B, and  D). A robust co-localization of BACE1 with p62 or ubiquitin was detected within dystrophic neurites surrounding amyloid plaques of hAPP mouse brains (Fig. 5E). 96.3% of p62 and 97.9% of ubiquitin co-localized with LC3-labeled AVs in mossy fibers enriched with axons and axon terminals of mutant hAPP Tg mice (Fig. 5, F and G). Consistent with cultured mutant hAPP neurons (Fig. 4), these observations in brain slice sections suggest autophagic retention of BACE1 in distal axons of AD neurons in vivo.
To determine whether aberrant distal retention of BACE1 impairs its delivery to the somatic lysosomes, we examined lysosomal targeting of BACE1 in the hippocampal regions of WT and hAPP mouse brains. We found that the fluorescence inten-sity of BACE1 in the soma was reduced to 71.28% in the hippocampal CA3 regions of mutant hAPP Tg mice compared with that of WT mice (p Ͻ 0.001) (Fig. 6, A and B). Furthermore, the co-localized mean intensity of BACE1 with LAMP-1, a lysosomal marker, was significantly reduced to 54.3% in hAPP mouse brains (p Ͻ 0.001) (Fig. 6, A and C). Thus, our data support the notion that AD-linked autophagic stress exacerbates BACE1 accumulation in distal axons, which may augment axonal pathology.
Autophagy-induced BACE1 Turnover Is Impaired in Mutant hAPP Tg Neurons-We next examined whether inducing autophagy in mutant hAPP Tg neurons alters BACE1 levels. Whereas autophagy induction enhances BACE1 turnover in WT neurons (Fig. 1), BACE1 levels in mutant hAPP neurons were maintained after a 24-h incubation with trehalose (1.16 Ϯ 0.08; p Ͼ 0.05). Substantial increase of BACE1 was detected when lysosomal function was suppressed by lysosomal inhibitors (1.59 Ϯ 0.13; p Ͻ 0.01) (Fig. 7, A and B), which suggests that lysosomal activity is unlikely impaired in mutant hAPP neurons. Consistently, trehalose treatment led to increased autophagic flux in mutant hAPP neurons, as evidenced by elevated levels of LC3-II (1.84 Ϯ 0.18; p ϭ 0.001). Further increase of LC3-II was detected in the presence of lysosomal inhibitors (2.81 Ϯ 0.41; p Ͻ 0.01) (Fig. 7, A and C). Treatment with   the second mTOR-independent autophagy inducer 10-NCP showed similar effects; autophagy-induced BACE1 turnover was abolished in mutant hAPP neurons (1.13 Ϯ 0.15; p Ͼ 0.05), but BACE1 levels were further increased in the presence of lysosomal inhibitors (1.63 Ϯ 0.14; p Ͻ 0.01) (Fig. 7, D and E). The levels of LC3-II were consistently elevated following a 24-h incubation with 10-NCP (2.25 Ϯ 0.24; p Ͻ 0.001; with lysosomal inhibitors, 3.58 Ϯ 0.53; p Ͻ 0.01) (Fig. 7, D and F). This result suggests that AV-associated reduction of AV retrograde transport disrupts autophagy-induced BACE1 turnover in mutant hAPP neurons. Consistently, elevated levels of C99 and C89 were also observed in mutant hAPP neurons following autophagy induction (Fig. 7A), suggesting an enhanced BACE1 processing of APP. To determine autophagic APP processing, we carried out co-immunostaining with MAP2 and 6E10 antibodies in mutant hAPP neurons following incubation with 10-NCP. GFP-LC3-labeled AVs were co-stained with 6E10marked APP or its cleaved products (␤-C-terminal fragments (CTFs) or A␤) along MAP2-negative axons (Fig. 7, G and H). Thus, our data indicate that increasing autophagic induction in AD neurons triggers autophagic retention of BACE1 and enhances BACE1 processing of APP.

Discussion
Autophagy is the major cellular degradation pathway, which consists of the following two key steps: 1) formation of double membrane autophagosomes engulfing misfolded proteins and damaged organelles to be degraded; 2) the fusion of the autophagosome with the lysosome into autolysosome, where autophagic cargoes are degraded (39 -41). Microtubule-based long-distance axonal transport is essential for autophagic clearance because autophagosomes are predominantly generated in distal axons and mainly rely on retrograde transport to somatic lysosomes for degradation (23,29,(31)(32)(33). Defective autophagy has been implicated in AD pathogenesis (27,42,43). The presence of massively accumulated AVs in dystrophic (swollen) neurites is a unique feature linked to AD pathology (38,44). These pathological features raise the following fundamental question. Does AD-associated autophagic stress interfere with BACE1 trafficking and thus augment BACE1 cleavage of APP in AD neurons?
Although many studies have been focused on BACE1 trafficking and turnover in non-neuronal cell types (16,45,46) or neuron-derived cell lines (12,15,47,48), its relevance to BACE1 long distance axonal transport and its retention and cleavage of APP in distal axons remains unknown. This study examined, for the first time, the autophagy-mediated regulation of BACE1 trafficking to lysosomes for degradation in WT and AD-related neurons. We show that BACE1 turnover is enhanced in primary WT cortical neurons upon autophagy induction but is suppressed when lysosomal function is inhibited (Fig. 1). Inducing autophagy triggers robust recruitment of BACE1 to the autophagy pathway; a significant portion of BACE1 co-localizes and co-migrates with AVs from distal axons toward the soma, thereby enhancing BACE1 trafficking to lysosomes (Figs.  2 and 3). To our surprise, this autophagy-induced BACE1 turnover is largely abolished in mutant hAPP neurons. Impaired retrograde transport results in aberrant accumulation of AVassociated BACE1 within the distal axons of neurons from ADrelated mutant hAPP J20 Tg mice (Fig. 4). Consistently, mutant hAPP Tg mouse brains recapitulate the phenotype of autophagic BACE1 retention in distal axons and reduced lysosomal targeting in the soma of the hippocampal regions (Figs. 5 and 6). Inducing autophagy in mutant hAPP neurons results in further accumulation of AV-associated BACE1, thereby augmenting ␤-cleavage of APP within AVs in distal axons (Figs. 7 and 8). Importantly, rescuing retrograde transport by overexpressing Snapin in mutant hAPP neurons reduces axonal retention of autophagic BACE1 and facilitates its targeting to lysosomes (Fig. 8). Therefore, we proposed that autophagy is one of the primary pathways regulating BACE1 trafficking and turnover FEBRUARY 3, 2017 • VOLUME 292 • NUMBER 5 in WT neurons. This pathway, however, is impaired under ADlinked autophagic stress, thus providing a new mechanistic insight as to why BACE1 is largely retained in distal axons, where ␤-cleavage of APP mainly occurs.

Regulation of BACE1 Degradation
Autophagosomes are predominantly generated in distal axons and then undergo predominant retrograde transport toward the soma (23,29,30,32,33). By recruiting LE-loaded dynein-Snapin (motor adaptor) transport complexes, nascent autophagosomes gain retrograde motility through fusion with LEs into amphisomes (32), which delivers autophagic cargoes to somatic lysosomes for clearance. LEs are enriched with amyloidogenic machinery and provide an optimal acidic environment necessary for ␤-secretase activity (10, 13-17, 20). BACE1 ultimately undergoes degradation within lysosomes (15,18,19). Thus, the traffic route for AV-associated BACE1 to lysosomes is critical for controlling its level and activity. Through autophagosome fusion with LEs into amphisomes, we showed that a significant portion of LE-loaded BACE1 is recruited to the autophagy pathway and co-migrates with AVs in the retrograde direction within axons upon autophagy induction. As a result, BACE1 utilizes AVs as a ride for its trafficking toward somatic lysosomes through dynein-Snapin-driven retrograde transport, which facilitates its turnover within lysosomes. Therefore, we proposed that autophagy serves as an alternative trafficking route to regulate BACE1 degradation in neurons.
Our current findings provide a mechanistic interpretation of many reported observations. Altered autophagy has been implicated in the pathogenesis of AD by its involvement in the showing that inducing autophagy augmented accumulation of AV-associated BACE1 in the axon of mutant hAPP neurons, which was reversed by Snapin-enhanced retrograde transport. Cultured cortical neurons were transfected with BACE1-GFP and mRFP-LC3 and HA-Snapin or HA vector, followed by a 24-h incubation with 100 mM trehalose (Tre) prior to time-lapse imaging at DIV 10 -12. Note that the axonal densities of both AVs and AV-associated BACE1 and the percentage of BACE1 associated with AVs were increased in mutant hAPP neurons following autophagy induction. The average numbers of AVs and BACE1 co-labeled with LC3 per 10-m axon were quantified in mutant hAPP neurons in the presence and absence of trehalose, respectively (B and C). The percentage of AV-associated BACE1 relative to total BACE1 was examined in the axon of hAPP neurons (D). E and F, quantitative analysis showing enhanced motility of AVs and AV-associated BACE1 in mutant hAPP axons expressing Snapin but not vector control upon autophagy induction. G and H, overexpressing Snapin facilitates BACE1 trafficking to lysosomes in the soma of mutant hAPP neurons. Co-localized mean intensity of BACE1 with LAMP-1 with and without trehalose treatment was examined. Data were quantified from at least four independent repeats and from a total number of AVs or AV-associated BACE1 (v) from a total number of neurons (n), as indicated in parentheses (E and formation of amyloid plaques (49,50). AVs are massively accumulated and clustered within dystrophic neurites of AD brains, and these AVs have been proposed as an alternative source for A␤ generation (38,49,51). Purified AVs from AD mice are enriched with A␤ peptide and its immediate precursor, ␤CTF, suggesting that BACE1 cleavage of APP occurs on the membrane of AVs (49,51). In this study, we show that a significant portion of BACE1 is recruited to the autophagy pathway in response to autophagy induction in healthy neurons (Figs. 1-3). However, under AD-linked autophagic stress with defective retrograde transport, we provide in vitro and in vivo evidence that AV-associated BACE1 is accumulated in distal axons along with impaired retrograde transport (Figs. 4 and 5). Under such pathological conditions, inducing autophagy exacerbated autophagic retention of BACE1 in distal axons of mutant hAPP neurons (Figs. 7 and 8). Thus, our study supports the notion that AD-associated autophagic stress retains BACE1 within the autophagy pathway and thus impairs its trafficking to lysosomes for degradation, thereby enhancing BACE1 processing of APP in AD axons.
Autophagy induction may represent a promising therapeutic strategy to prevent or treat AD and other neurodegenerative diseases. Studies have shown the beneficial effects of autophagy activation on reducing the accumulation of polyglutamine expansions and ␣-synuclein and on ameliorating disease-associated pathology (52)(53)(54). However, the role of autophagy in AD is controversial (55)(56)(57)(58)(59)(60)(61). Several studies reported that increasing autophagy showed the rescue effects when AD mice were administered rapamycin prior to the development of AD pathology (58,60,61) but had no effect on mice with already established AD-like pathology and cognitive deficits (60). Autophagy activation has also been shown to trigger an increase in A␤ levels and is involved in amyloidogenesis (49,50). Consistent with these findings, we proposed that impaired retrograde transport in AD axons compromises beneficial effects of increasing autophagy activation on AD pathology by exacerbating autophagic accumulation of BACE1 and augmenting ␤-cleavage of APP and axonal pathology. Therefore, our study establishes a foundation for future investigation into the potential therapeutic strategy of increasing autophagy induction at the early disease stages.

Experimental Procedures
Mice-hAPP mice (C57BL/6J) from line J20 were purchased from The Jackson Laboratory. The platelet-derived growth factor (PDGF) ␤ chain promoter was used to direct neuronal expression of the familial AD-linked human APP Swedish and Indiana mutants in this J20 mouse line (62).
Transfection and Immunocytochemistry of Cultured Cortical Neurons-Cortices were dissected from E18 to E19 mouse embryos as described (22,64,65). Cortical neurons were dissociated by papain (Worthington) and plated at a density of 100,000 cells per cm 2 on polyornithine-and fibronectin-coated coverslips. Neurons were grown overnight in plating medium (5% FBS, insulin, glutamate, G5, and 1ϫ B27) supplemented with 100ϫ L-glutamine in Neurobasal medium (Invitrogen). Starting at DIV 2, cultures were maintained in conditioned medium with half-feed changes of neuronal feed (1ϫ B27 in Neurobasal medium) every 3 days. Primary hAPP Tg neurons were cultured from breeding mice of hemizygous mutant hAPPSwe/Ind Tg (J20 line) with WT animals (62). Genotyping assays were performed following culture plating to verify mouse genotypes. In our study, we examined both transgenic neurons and non-transgenic neurons derived from their littermates. Neurons were transfected with various constructs using Lipofectamine 2000 (Invitrogen) followed by time-lapse imaging at DIV 10 -12 transfection prior to qualification analysis.
For immunostaining, cultured neurons were fixed with 100% ice-cold methanol at Ϫ20°C for 10 min, washed three times with 1ϫ PBS for 5 min each, and then incubated in 0.4% saponin, 5% normal goat serum, and 2% bovine serum albumin (BSA) in PBS for 1 h. Fixed cultures were incubated with primary antibodies in PBS with 2% BSA and 0.4% saponin at 4°C overnight. Cells were washed four times with PBS at room temperature for 5 min each, incubated with secondary fluorescent antibodies at 1:400 dilution in PBS with 2% BSA and 0.4% saponin for 30 min, re-washed with PBS, and then mounted with Fluor-Gel anti-fade mounting medium (EMS) for imaging.
For semi-quantitative analysis, protein bands detected by ECL were scanned into Adobe Photoshop CS6 and analyzed using ImageJ (National Institutes of Health). Care was taken during exposure of the ECL film to ensure that intensity readouts were in a linear range of standard curve blot detected by the same antibody. Paired Student's t tests were carried out, and results are expressed as mean Ϯ S.E.
Tissue Preparation and Immunohistochemistry-Animals were anesthetized with 2.5% avertin (0.5 ml per mouse) and transcardially perfused with fixation buffer (4% paraformaldehyde in PBS, pH 7.4). Brains were dissected out, postfixed in fixation buffer overnight, and then placed in 30% sucrose at 4°C. The 10-m-thick coronal sections were collected consecutively to the level of the hippocampus and used to study co-localization of various markers. After incubation with blocking buffer (2.5% goat serum, 0.15% Triton X-100, 1.5% BSA, 0.5% glycine in H 2 O) at room temperature for 1 h, the sections were incubated with primary antibodies at 4°C overnight, followed by incubating with secondary fluorescence antibodies at 1:400 dilution at room temperature for 1 h. After fluorescence immunolabeling, the sections were stained with DAPI and washed three times in PBS. The sections were then mounted with antifading medium (Vector Laboratories, H-5000) for imaging. Confocal images were obtained using an Olympus FV1000 oil immersion ϫ40 objective with sequential acquisition setting.
Eight to 10 sections were taken from top-to-bottom of the specimen, and the brightest point projections were made.
Quantification of Co-localization-A threshold intensity was determined in the thresholding function of ImageJ, which was preset for the fluorescent signals as described previously (66). The co-localized pixels above the threshold intensity were automatically quantified and scored by ImageJ based on the fluorescence intensity profile, which was expressed as co-localized mean intensity positive for both channels. The co-localization was presented as the percentage of the co-localized intensity relative to total fluorescence intensity.
Image Acquisition and Quantification-Confocal images were obtained using an Olympus FV1000 oil immersion ϫ60 objective (1.3 numerical aperture) with sequential acquisition setting. For fluorescent quantification, images were acquired using the same settings below saturation at a resolution of 1024 ϫ 1024 pixels (8-bit). Eight to 10 sections were taken from top-to-bottom of the specimen, and the brightest point projections were made. Morphometric measurements were performed using ImageJ. Measured data were imported into Excel software for analysis. The thresholds in all images were set to similar levels. Data were obtained from at least three independent experiments, and the number of cells or imaging sections used for quantification is indicated in the figures. All statistical analyses were performed using the Student's t test and are presented as mean Ϯ S.E.
For live cell imaging, cells were transferred to Tyrode's solution containing 10 mM Hepes, 10 mM glucose, 1.2 mM CaCl 2 , 1.2 mM MgCl 2 , 3 mM KCl, and 145 mM NaCl, pH 7.4. Temperature was maintained at 37°C with an air stream incubator. Cells were visualized with a ϫ60 oil immersion lens (1.3 numerical aperture) on an Olympus FV1000 confocal microscope, using 488 nm excitation for GFP and 543 nm for mRFP. Time-lapse sequences of 1024 ϫ 1024 pixels (8-bit) were collected at 1-2-s intervals with 1% intensity of the argon laser and maximum pinhole opening to minimize laser-induced bleaching and damage to cells. Dual color time-lapse images were captured by a total of 100 frames. All recordings started 6 min after the coverslip was placed in the chamber. The stacks of representative images were imported into ImageJ. A membranous organelle was considered stopped if it remained stationary for the entire recording period; a motile one was counted only if the displacement was at least 5 m.
Criteria for Axon Selection in Cultured Neurons-For analyzing the motility of AVs or late endosomes in live neurons, we selected axons for time-lapse imaging and measuring motility because axons, but not dendrites, have a uniform microtubule organization and polarity. Axonal processes were selected as we reported previously (22,64,67). Briefly, axons in live images were distinguished from dendrites based on the following known morphological characteristics: greater length, thin and uniform diameter, and sparse branching (68). Only those that appeared to be single axons and separate from other processes in the field were chosen for recording axonal transport of AVs or BACE1 vesicles. Regions where crossing or fasciculation occurred were excluded from analysis.
Kymographs were used to trace axonal anterograde or retrograde movement of membranous organelles and to count sta-tionary ones as described previously (67,69) with extra plugins for ImageJ. Briefly, we used the "Straighten" plugin to straighten curved axons and the "Grouped ZProjector" to z-axially project re-sliced time-lapse images. The height of the kymographs represents recording time (100 s unless otherwise noted), whereas the width represents the length (m) of the axon imaged. Counts were averaged from 100 frames for each time-lapse image to ensure accuracy of stationary and motile events. Measurements are presented as mean Ϯ S.E. Statistical analyses were performed using unpaired Student's t tests.
Author Contributions-T. F., P. T., and Q. C. designed the project; T. F., P. T., C. A., Y. Y. J., and Q. C. performed experiments and analyzed data; Q. C. wrote the manuscript.